A fast source camera identification and verification method based on PRNU analysis for use in video forensic investigations

Due to the rapid development of digital and cloud technologies, everyone can easily shoot and spread digital videos via email or social media. However, it is difficult for law enforcement to trace the origin of those digital videos, while some videos or images containing illegal information such as personal privacy, obscene pornography, and national security-related content. Recently, a significant breakthrough is achieved by using Photo-Response Non-Uniformity (PRNU) noise to characterize the camera sensor. However, PRNU analysis is often carried out on a frame-by-frame basis. As a result, the processing time is unbearable when treating a large set of videos and devices. In this paper, we propose a novel video forensic method considering both cameras rolling and I-frame of videos to improve the processing time and accuracy. Experimental results demonstrate that our proposed method is at a minimum of 15 times on average faster than the most wildly used method, PRNU analysis, and reduce the false positive rate as compared to existing methods used in the field of the forensic examination.